Abstract Rationale Bronchopulmonary dysplasia (BPD) is the second most common chronic pediatric respiratory disease - affecting approximately 35% of preterm infants. The development of BPD is multifactorial, with emerging evidence highlighting the contribution of social determinants of health (SDoH) to BPD risk. While there is growing evidence supporting specific demographic and clinical factors that may predict disease, investigations relating SDoH and BPD development have been understudied. We aimed to determine how the addition of SDoH impacts an existing risk estimation tool for BPD severity or death, using readily available sociodemographic and clinical data. Methods We utilized the Yale Patterns of Growth and Outcomes (POGO) database. In addition to electronic medical records (EMR) review conducted by study personnel to obtain birth and health information, parents also completed sociodemographic surveys regarding insecurities ( food, housing, income, etc.) from which a binary SDoH variable was determined. BPD was classified and defined using the Jensen 2019 criteria. Multivariable logistic regression models were used to analyze factors associated with BPD. As the objective of the study was prediction, forward stepwise covariate selection was used, with p-value 0.2 as the threshold for model entry. Final models were selected after the exclusion of covariates with p-value 0.01. Results We analyzed 312 patients with complete data: 209 infants with No BPD, 40 infants with Grade 1 BPD, 45 infants with Grade 2 BPD, 18 infants with Grade 3 BPD or Death. 236 infants had reported SDOH risk and 76 infants did not have SDoH risk. In the univariate analysis, gestational age, birth weight, and race group were associated with BPD outcomes. Forward stepwise regression resulted in no covariates included in the final model. After comparing the effect of ethnicity, race group, and SDoH on model fit, only the inclusion of race group improved model fit (as measured by Akaike Information Criterion) for the outcome of binary BPD, BPD Jensen categories, and BPD collapsed outcomes (Grade 3 BPD and death were combined). Conclusion Contrary to our hypothesis, SDoH did not improve model fit for the prediction of BPD. Future studies would benefit from larger sample size and corresponding disaggregation of race and ethnicity groups and we plan to further utilize this data to investigate the effect of race, ethnicity, and SDoH on BPD discordance. This abstract is funded by: Dr. Sarah Taylor was partially supported by grants from the National Institutes of Health (NIH R01 HD113734, NIH R01 HD108646, NIH R01 HD106359, NIH R01 HD111633, NIH R01 HD112396.) Dr. Samuel Gentle was partially supported by a grant from the Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health (K23HD113837-01A1). Dr. Ruthfirst Ayande was partially supported by a research grant from The Nestle Foundation. Mallory Go was partially supported by a Research at Brown grant from Brown University.
Go et al. (Fri,) studied this question.